Hi list,
I'm expiriencig some yet-to-be-determined problems using xfs with nfs,
on two different boxen in two different setups.
1. Redhat71, 2.4.2-2+xfs kernel and other relevant xfs RPMs from
oss.sgi.com ftp; two 180gb scsi disks in raid1; stock rh71 nfs-utils
rpm. Client is solaris7. Mount works ok, copying files back and forth
works ok, then, after some time, nfs dies ('nfs server not responding'
in solaris logs) altough there are no error msgs in redhat logs.
Restarting nfs service on redhat does not make any difference for the
current stalled mount on solaris.
2. Redhat71, 2.4.5-0.2.9+xfs rawhide kernel and other relevant xfs RPMs;
raid5 array of 8 ide disks hanging on 3ware card; stock rh71 nfs-utils
rpm. Client is OpenBSD 2.9. Mount works (v3, udp), for testing i tried
to copy a whole openbsd cvs tree to the nfs server. About 220mb was
copied over, then the server hung; couldn't ssh to it, didn't respond to
console input, but ping worked. I rebooted it, exported an ext2
partition, mounted it on openbsd and succesfully copied cvs tree back
and forth a couple of times. Then I exported the xfs partition again,
tried the same copy, got another lockup. No msgs in the logs, no kdb,
nothing.
Both servers are using intel NICs with intel's e100 v1.6.6 driver.
While the second setup always near me, I can only access first setup via
ssh.
I know that people are running xfs with success on production
fileservers, so i'd prefer to think it's me doing something wrong. But
both setups are basic stuff, without any fancy nfs export or mount
options. I'd like to figure out what exactly is the problem here and
who's to blame :)
Just finishing the mail, something popped up on the nfs server's console
here:
Kernel panic: kmem_zone_zalloc: NULL memory on KM_SLEEP request!
... that would be?
Thanks for any help.
--
Jure Pecar
|